Reply to thread

Message

<blockquote data-quote="Inker" data-source="post: 645530" data-attributes="member: 10541">Re: AW: Trouble with ImportingWell, last I checked, the processing is still very much sequential, much more so than I would like personally, but anyways. So unless you received timeouts with an older version nothing should have really changed. Normally the worst thing that can happen is that metadata for that particular series is not available until the next run (for instance when mp is run). Now its possible that if a timeout happens this exception propagates to the ui thread in a way that it doesn't know how to deal with. That is how I understand your post and I agree. However, tesla has no such timeouts in his log, this would not have anything to do with the amount of logging going on, and the fact that MSE could cause this is if that was the reason for the timeouts.My only guess left is that the way we do logging is causing MSE to freak out, which causes the file to be locked, which causes us to wait on it, all the while multiple threads try to do their work blocking on it on hammer on the lock, and that we have a bug that causes a deadlock in such a scenario. We could test this by never releasing our lock on the log file for the entire duration of our run. I will try to find some time/motivation over the next few days. However it would seriously help if someone who can reproduce this were on IRC.</blockquote>

[QUOTE="Inker, post: 645530, member: 10541"] [b]Re: AW: Trouble with Importing[/b] Well, last I checked, the processing is still very much sequential, much more so than I would like personally, but anyways. So unless you received timeouts with an older version nothing should have really changed. Normally the worst thing that can happen is that metadata for that particular series is not available until the next run (for instance when mp is run). Now its possible that if a timeout happens this exception propagates to the ui thread in a way that it doesn't know how to deal with. That is how I understand your post and I agree. However, tesla has no such timeouts in his log, this would not have anything to do with the amount of logging going on, and the fact that MSE could cause this is if that was the reason for the timeouts. My only guess left is that the way we do logging is causing MSE to freak out, which causes the file to be locked, which causes us to wait on it, all the while multiple threads try to do their work blocking on it on hammer on the lock, and that we have a bug that causes a deadlock in such a scenario. We could test this by never releasing our lock on the log file for the entire duration of our run. I will try to find some time/motivation over the next few days. However it would seriously help if someone who can reproduce this were on IRC. [/QUOTE]

Verification